Mathematical center in Akademgorodok invites you to take part in the seminar on May 29 at 6 pm Novosibirsk time (UTC+7).
The seminar鈥檚 heads 鈥� I.A. Taimanov, S.I. Kabanikhin, A.E. Mironov, M.A. Shishlenin.
The secretary 鈥� N.M. Prokhoshin (
n.prokhoshin@g.nsu.ru).
I. V. Oseledets (Skolkovo Institute of Science and Technology) will present his talk
鈥淢odern challenges of machine learning and big data鈥�.
The talk will describe several major research areas in machine learning and their application in mathematical modeling and inverse problems鈥� solutions.
Deep learning methods have let us reach outstanding results in various spheres, such as image and text recognition. However, there is a major weakness in quality assessment for the obtained models in terms of reliability, stability and malicious attacks. Currently there are no stable architectures and approaches, moreover, the known estimates are too 鈥渨eak鈥�. Having said all that, usage of machine learning in modeling still grows rapidly.
We will discuss the theory about fundamental ability to estimate solutions of different classes of equations using deep neural networks. And we will also cover algorithms of the construction of such approximations.
To take part, please, on May 29 after 5:45 pm Novosibirsk time (UTC+7) connect to the Zoom conference via this link: .
You can also connect manually in the Zoom app using the Conference ID: 977 2926 2391.
Please pay attention to the following rules of attending the online seminar:
- As a security measure, it is highly advisable to connect using your real name.
- You are free to join the discussion after the end of the report. However, during the report please keep you microphone turned off.
- For the convenience of the speaker, we recommend writing your questions in the chat in advance.
